Split `ThreeFlavorDecoherence` into “black box” and `in_sn` transformations

[JostMigenda]

I noticed that ThreeFlavorDecoherence is described as Phenomenological transformation and placed in __init__.py. It really should be a SN transformation. It's true that Earth matter can't change the decoherence but vacuum transformations could e.g. one of the mass states decays.

My impression was that NoTransformation, CompleteExchange and ThreeFlavorDecoherence are typically used as “black box transformations”—where researchers are interested in some other aspect and use these extreme scenarios to get a rough idea of how flavor transformations could affect studies of that other aspect. So it makes sense to me to keep these three separate?

However, if you think that any users might want to build a TransformationChain out of ThreeFlavorDecoherence in the SN plus additional transformations in vacuum/earth, then we could turn TFD into a SNTransformation. In that case, would TransformationChain(in_sn=ThreeFlavorDecoherence()) automatically result in the same flavour equilibration that we get from the current implementation of TFD? If not, that might be very confusing.

The use case you describe for TFD is a useful one so my suggestion is to have a Phenomenological Transformation in init.py called e.g. Equilibrate which implements this case, and to have ThreeFlavorDecoherence as prescription in in_sn.py. The ThreeFlavorDecoherence is useful for describing the effect of strong turbulence in supernovae; the TwoFlavorDecoherence describes the effect of weak turbulence.

Originally posted by @jpkneller in #351 (comment)